victoraut/liken-skills
Agent skills for Liken: near-deduplication and record linkage for Python DataFrames.
78
Quality
98%
Does it follow best practices?
Impact
—
No eval scenarios have been run
Securityby
Passed
No known issues
README.md
liken-skills
Agent skills for Liken — a Python library for near-deduplication, fuzzy matching, and record linkage over pandas, Polars, Modin, Dask, Ray and PySpark DataFrames.
The skills
| Skill | API tier | Teaches |
|---|---|---|
liken | overview / router | What Liken is, the simplest exact dedupe, and which skill/API to reach for. |
liken-dedupers | apply dedupers | Built-in dedupers (exact, fuzzy, tfidf, lsh, jaccard, cosine + predicates), .apply() with a single deduper or a dict collection, pandas affordances. |
liken-pipelines | pipelines | lk.pipeline()/lk.col(), AND/OR/NOT semantics, preprocessor scoping, rule predication. |
liken-custom-dedupers | extension | Defining and registering custom dedupers with @lk.custom.register. |
liken-record-linkage | entity resolution | .canonicalize(), .collect(), .canonicals(), .synthesize(). |
liken-backends-performance | scaling | Backend selection/extras, spark_session, blocking keys, LSH, cost model. |
Install
tessl install liken/liken-skills@0.1.0Links
- Library docs: https://victoraut.github.io/liken/
- Repository: https://github.com/VictorAut/liken